Electroabsorptive
modulated lasers bolster long-haul and DWDM use

BY CHARLES JACKSON,
RAY NERING, and ANDY ZHOU
Lucent Technologies
Breinigsville, PA

The telecommunications industry's continued rapid growth is being fueled by a technology that allows system capacity to keep pace with user demand. Dense-wavelength-division-multiplexed (DWDM) systems hold substantial economic appeal to network operators because they increase that capacity by adding wavelengths to already-installed bases.

Electroabsorptive modulated lasers are now
capable of supporting span lengths that exceed
1,000 km for long-haul and DWDM applications.

And the laser of choice for high-bandwidth DWDM systems is rapidly becoming the electroabsorptive modulated laser (EML). These lasers have also been successfully deployed in public networks worldwide. Currently, they are produced at standard wavelengths that are specified by the International Telecommunications Union (ITU), which promotes the availability of passive DWDM components.

EMLs integrate a laser and modulator into a single chip and are coupled to a single-mode fiber in an industry-standard 14-pin butterfly package. The laser section of an EML chip is operated in a continuous-wave (CW) mode. The input signal is applied to a modulator that turns the light on and off, thereby producing a very low chirp signal that can be propagated through very long lengths of standard embedded fiber with minimal signal degradation.

EMLs for the long haul

EMLs are typically used in tran-sponders for long-haul systems. The transponder takes a SONET OC-48 signal, which usually originates from inside the same central-office facility at 1.3 or 1.5 µm and converts that signal to a specific wavelength (see Fig. 1 ). By combining many incoming signal streams to a variety of ITU channels, the capacity of a single fiber can be easily increased by a factor of 8, 16, or more, without replacing all of the equipment in the central office.

Fig. 1. Transponders typically take a SONET OC-48 signal originating
from inside the same central-office facility
at 1.3 or 1.5 µm, and convert the signal to a specific wavelength
as specified by the ITU.

EMLs are now capable of supporting span lengths in excess of 1,000 km, which is far beyond the 200-km range offered by lasers that are typically operated under direct modulation. These lasers use distributed feedback (DFB), and emit virtually all their light at a single wavelength.

During direct modulation–when the drive current is modulated in proportion to the input signal–the change in laser current affects the optical characteristics of the laser cavity. This change results in slight instantaneous changes in laser wavelength and is commonly referred to as chirp. As a result, the signal emitted from the EML is distributed over a changing narrow range of frequencies.

As the light pulse travels down the fiber, it becomes distorted because of the chromatic dispersion properties of the fiber. This phenomenon occurs because light of different wavelengths propagates at different speeds. If the distortion of the pulse becomes significant, the sensitivity of the receiver will be affected.

The difference between the sensitivity with and without fiber is called the dispersion penalty. Generally, lasers that have a dispersion penalty of less than 2 dB over 100 km are considered acceptable for telecommunications applications.

Because the laser portion of an EML is operated in CW mode, there is no broadening of the spectral width of the light signal due to modulation of the laser current. This enables the light signal to travel through longer lengths of fiber with less distortion than direct-modulated devices. Although laser current in an EML is held constant, a relatively small amount of chirp generated by the modulator eventually limits the useful distance over which the device can be used.

Because all EMLs are characterized for chirp, time-resolved spectroscopy measurements are used to determine how their wavelengths shift with respect to the input signal. The EML chirp–characterized as peak-to-peak chirp–is the maximum wavelength shift during the transition of the modulator from a high or transparent state to an off or attenuating state, and vice versa.

The peak-to-peak chirp is critical in determining whether a device can be used for long distances. In 640-km applications, devices that have a peak-to-peak chirp of less than 0.15 Å (modulator bias is equal to 0 V) are acceptable.

EML design goals for

DWDM applications

Since DWDM system design plays a very significant role in overall link performance, several factors need to be considered:

• How many optical amplifiers should be used in the system?

• What is the power level in the fiber?

• How many wavelengths will be running in the system?

• Should allowances be made for bandwidth upgrade strategy?

Lucent Technologies prefers to measure the chirp of the EML independently of the system design because the chirp can be affected by several variables. The nonlinear effects of the fiber, noise from the multiple optical amplifiers, crosstalk of the optical multiplexer and demultiplexer, multiwave-mixing, and self-phase modulation are only a few of the factors that influence overall system performance.

The same laser can have a variety of performance characteristics, depending on system configuration. As such, system designers benefit most by being provided with EML chirp performance, as opposed to dispersion-penalty performance, which is highly system dependent.

Wavelength stability and aging

Wavelength stability is extremely critical in DWDM systems where a change of 5 Å can place one channel on top of another. As lasers age, their wavelengths drift toward the blue part of the spectrum, suggesting that if they were tuned toward the red side of a channel there would be more margin for aging than if they were centered in the channel band.

Individual EMLs in a typical DWDM system must be tuned to a particular wavelength, which is accomplished by changing the chip temperature. EML packaging is very similar to other laser packages in that the EML chip is mounted on a platform in conjunction with a thermistor and a back-facet monitor.

The platform is mounted on top of a thermoelectric cooler inside a hermetic package. By driving current into the cooler and using the thermistor for feedback, the temperature of the EML chip can be stabilized within a specific range.

The wavelength temperature dependence of an EML is typically 0.08 nm/°C, which is the same as for standard DFB lasers. Nominal operating temperature is 25°C for EMLs, which are usually tuned to a particular wavelength by changing the chip temperature by 15° to 35°C.

Also, when determining the wavelength temperature tuning range, maximum operating temperature becomes very important. The maximum differential temperature of the thermoelectric cooler is about 50°C, which means that if the maximum case temperature is 65°C, the minimum tuning temperature is 15°C. If higher-temperature operation is required, then the tuning will have to be adjusted.

The drive current of the laser will also affect the wavelength of the device. Although this effect is an order of magnitude lower than the temperature effect, it still must be taken into account when all of the channels are being balanced in the system.